An Improved Deadbeat Predictive Current Control of PMSM Drives Based on the Ultra-local Model

Abstract

Deadbeat predictive current control (DPCC) has been widely applied in permanent magnet synchronous motor (PMSM) drives due to its fast dynamic response and good steady-state performance. However, the control accuracy of DPCC is dependent on the machine parameters' accuracy. In practical applications, the machine parameters may vary with working conditions due to temperature, saturation, skin effect, and so on. As a result, the performance of DPCC may degrade when there are parameter mismatches between the actual value and the one used in the controller. To solve the problem of parameter dependence for DPCC, this study proposes an improved model-free predictive current control method for PMSM drives. The accurate model of the PMSM is replaced by a first-order ultra-local model. This model is dynamically updated by online estimation of the gain of the input voltage and the other parts describing the system dynamics. After obtaining this ultra-local model from the information on the measured stator currents and applied stator voltages in past control periods, the reference voltage value can be calculated based on the principle of DPCC, which is subsequently synthesized by space vector modulation (SVM). This method is compared with conventional DPCC and field-oriented control (FOC), and its superiority is verified by the presented experimental results.